Regression in detection of a copyright statement

[fviernau]

Setup

• Scancode version 32.5.0, or pick a version which reproduces from below table.
• Ubuntu 24

Reproduce

1. Scan /seed_local.h with options -c -l --json-pp out.json
2. Note that detection in line 11 is

        {
          "copyright": "Copyright (c) 2007",
          "start_line": 11,
          "end_line": 11
        }

3. But expected: Copyright (c) 2007 KISA(Korea Information Security Agency).

Scancode versions

Version   Extracted copyright
--------  -------------------------------------------
31.0.2    Copyright (c) 2007 KISA(Korea Information Security Agency)
31.1.1    Copyright (c) 2007 KISA(Korea Information Security Agency)
31.2.6    Copyright (c) 2007 KISA(Korea Information Security Agency)
32.0.8    Copyright (c) 2007
32.1.0    Copyright (c) 2007
32.2.1    Copyright (c) 2007
32.3.1    Copyright (c) 2007
32.3.2    Copyright (c) 2007
32.3.3    Copyright (c) 2007
32.4.0    Copyright (c) 2007
32.4.1    Copyright (c) 2007
32.5.0    Copyright (c) 2007

[pombredanne]

@fviernau Thanks!

[pombredanne]

Git bisect to the rescue!

$ cat cop-check.sh 
./configure --clean
./configure --dev
scancode -q -c --yaml - cop.txt | grep -F "copyright: Copyright (c) 2007 KISA(Korea Information Security Agency)"
$ chmod u+x cop-check.sh
$ git bisect start
$ git bisect bad v32.0.8
$ git bisect good v31.2.6
warning: refname 'v31.2.6' is ambiguous.
Bisecting: 413 revisions left to test after this (roughly 9 steps)
[1fa4668ad8a9579a43782672f3cf895130501bc8] Fix scancode_version bug
$ git bisect run ./cop-check.sh
....
[600a00f80799aa446c3315cd89dc22b102d039ad] Improve copyright detection
....
600a00f80799aa446c3315cd89dc22b102d039ad is the first bad commit
commit 600a00f80799aa446c3315cd89dc22b102d039ad
Author: Philippe Ombredanne <pombredanne@nexb.com>
Date:   Fri May 12 01:23:00 2023 +0200

    Improve chttps://github.com/aboutcode-org/scancode-toolkit/commit/600a00f80799aa446c3315cd89dc22b102d039ad#diff-68f1966d414a58a3a75ae92a79baef800e2085b0b95404b0911483a0b022929dR1063opyright detection
    
    Also update some license ignorables and add new SAX license rules
    
    Signed-off-by: Philippe Ombredanne <pombredanne@nexb.com>

So the culprit commit is that one 600a00f

And checking the diff, that bit 600a00f#diff-68f1966d414a58a3a75ae92a79baef800e2085b0b95404b0911483a0b022929dR1063

+    # parens in the middle: for(var
+    (r"^[a-zA-Z]+[\)\(]+,?[\)\(]?[a-zA-Z]+[\.,]?$", 'JUNK'),

[tsteenbe]

@pombredanne I working to getting you more test cases on where saw copyright detection regression. Do you still need those or the root cause clear and a fix can be implemented?

[pombredanne]

@tsteenbe re:

I working to getting you more test cases on where saw copyright detection regression. Do you still need those or the root cause clear and a fix can be implemented?

Please ... more regressions :)